Robo-Trucks: Could they be first?
Last week, I attended The Information’s AV Summit in San Francisco, CA, which is one of the more intimate but high-visibility autonomous technology event in the space. This year, the spotlight shifted from robo-taxis to robo-trucks. It seems that until now, the industry somehow had missed the more straightforward use case and growing multibillion-dollar market. At last, technology companies and venture capitalists have overcome their car-centered fever. They are now listening, investing and trailing ahead, pun intended.
I was surprised when I heard the statistics clouding the freight industry. Every 15 minutes, a person is killed or seriously injured in an accident caused by a tractor trailer (a.k.a. 18-wheelers or semi-trucks). Around 500,000 trucking accidents occur each year in the US, with 1% of those resulting in death. There’s definitely room for improvement. As already seen with ADAS enhancing vehicle safety performance, robo-trucks are promising to spare more truckers’ and drivers’ lives.
Additionally, the technology brings other benefits around truck utilization. Because the central computer would follow best driving practices, maintenance would be reduced while fuel efficiency would increase. Also, with a non-stop twenty-four hour driving shift, cargo would arrive earlier to its destination.
It’s a no-brainer. Autonomous trucks (AT) won’t need to worry about little Billy biking on the intersection or fluffy Rufus running off to the middle of the street. Long haul trucking takes place for the most part on highways that expand miles and miles into eternity, mostly on a straight path. There are just cars and other trucks driving along. No pedestrians; no stop signs, no traffic lights. It’s like AT heaven for self-driving startups.
Robo-trucks, or ATs, will be in the business of transporting goods and not people. In principle, there won’t be someone inside the cabin whose life must be protected (in a Level 4 automation scenario). Similarly, ATs will operate within more predictable operational design domains. These factors make the technology stack less complex to develop, test and validate. For instance, maneuvering would only entail switching and merging lanes. There won’t be unprotected left turns involved or yielding to others at a stop sign.
Trucks will drive at specific highway speeds rather than at fluctuating inner city speeds. They will be able to drive at late hours of the night when there’s no traffic, as everyone else is at home sleeping. Current truck drivers are obligued by law to rest after eleven hours (raising the economic attractiveness of self-driving trucks). In contrast, robo-taxis will need to operate at peak times when people are getting to and from work within the metro area. This is the core commercialization strategy for light-duty vehicles, to have high utilization rates to maximize revenue per day.
The intelligent computer inside a self-driving car will need to account for congested city traffic and unpredictable scenarios for which algorithm models may not have been trained for. Therefore, solving the perception, prediction and planning for driverless ridesharing is ten times harder than for trucks.
Nevertheless, not all is rosy in the land of automated long-haul freight. Eighteen-wheelers still need to be aware of their surroundings, and be ready to react if vehicles around them suddenly de-accelerate. For instance, due to their size and heavy cargo, trucks carry more momentum (a function of mass x velocity) than their light-duty peers. A fully loaded truck can weigh up to 80,000 lbs while a car weighs only 3,500 lbs. So a truck cruising at 60 mph down Interstate-95 will need longer stopping distance than a car traveling at the same speed. We have the laws of physics to thank for this.
Today, the trucking industry leverages the decades of experience of human truckers on the highway. The challenge for self-driving truck companies will be to substitute those twenty years plus of skilled driving with intelligent algorithms. Even, make the latter better ‘at the wheel’. Consequently, we see strong initiatives behind the sophisticated truck technology underscoring the importance of safety in design engineering. The conversation is about safety critical elements, failure mode & effect analysis, and a system’s engineering approach.
In terms of logistics, the self-driving truck would cover the long-haul portion of the trip, whereas the first and last mile would still need to be driven by a human operator. For such setup to work out, truck fleets would have to operate under a transfer model. The self-driving trucks would be ‘geo-fenced’ to travel from exit to exit on the highway and drop off their trailer at depot stations for truckers to do the inner city driving. Similarly, there are other activities that technology companies haven’t figure out for self-driving trucks; such as who’d fuel the vehicle? The infrastructure so far hasn’t offered yet an autonomous pumping station…perhaps in the works?
Another question that companies are busy solving is the monetization of the trucking technology. For robo-cars, Uber has already setup an initial framework: smart phone app + ridesharing => carless urban dwellers. Even you and I can relate to this as it’s been disseminated ubiquitously. But when you go from a business to customer (B2C) to a business to business (B2B) market strategy, the segment target becomes more ambiguous. Who’s your customer? What is your main distribution channel?
Like most robo-taxi companies, which source their vehicle platform from OEMs and retrofit it with sensors (except for Zoox), self-driving companies are using Peterbilt-made trucks and focusing exclusively on the software. Except for OEMs such as Tesla, Daimler, and Volvo, not many other players in the industry are taking the vertically-integrated approach for self-driving trucks.
Startups like TuSimple plan on offering their self-driving trucks for purchase to fleet managers via their OEM partners. The self-driving company would support operations based on a subscription model. Embark, the blue truck ‘tech company’ (as 23-year old CEO Alex Rodrigues describes it) will have its system be factory-installed by OEMs rather than offer it as an aftermarket kit. Nevertheless, the company doesn’t discount a future where it licenses its self-driving software to truck operators.
But self-driving truck teams are not solely focusing on perfecting the technology. They’re already generating revenue. Many of these savvy startups are accumulating miles driven on highways transversing from the East to the West Coast. They’ve locked in partnerships to transport cargo for companies such as Amazon. Bigger contenders like Waymo are using their trucks to transport freight between Google data centers. It seems commercialization is coming at a much faster rate in the heavy-duty class than for passenger-focused robo-taxis.
You’d think that by now with all the technology potential, we’d have robo-trucks swarming our roadways, nationwide. However, since trucking is a much more conservative, robust, and traditional industry, compared to ridesharing, its legacy represents a gargantuan beast to tame.
The biggest hurdle today is regulation. Across the US, states have different laws about self-driving trucks. In contrast with robot-taxis, which operate within an urban area, semis cross state lines daily. And while places like Nevada, Florida, and Michigan have robo-truck friendly rules, others do not share the same welcoming attitude. Unfortunately, at federal level the policy doesn’t get any more progressive. The inaction by regulators may also have to do with the push back from truckers’ unions, as they see their jobs at risk.
Similarly, there’re remaining technology issues when coping with the real world. This is true for highways in northern terrritories; where it snows heavily for at least three months out of the year. Until the cameras and Lidar sensors don’t get better in harsh weather conditions, we’ll continue seeing these shiny red, blue, white and green trucks chasing the sun in the lower bottom half of the country.
Robo-trucks are coming. Actually, they already went, came, and turned back around again. Companies are racing each other to clock in more miles and cargo, when in fact there’s room for everyone to share in the profits. If the trucking industry were a country and the 700 billion in annual revenue was its GDP, it’d rank 34th, after Vietnam.
***
I hope you enjoyed this post related to my book, Autonomousity: Autonomous Vehicles & Emerging Business Models. You can check it out via this link: https://www.amazon.com/dp/B07QDM7HTX
I’d love to continue the conversation about self-driving cars! You can either leave a comment in my Medium page or connect with me via email at BejaranoAPaula@gmail.com or LinkedIn.
And don’t miss our Self-Driving 101 AV Tech Stack Panel Discussion on July 25th in San Francisco.
